The implications of three major new trials for the effect of water, sanitation and hygiene on childhood diarrhea and stunting: a consensus statement

We judge these trials to have high internal validity (Fig. 1; full table in Additional file 1), constituting good evidence that these specific interventions – as delivered in these settings – had no effect on childhood linear growth, and mixed effects on childhood diarrhea. Our view is that fidelity and compliance were at least as high as what might reasonably be expected in a typical WASH project or program.

In all three trials, these basic WASH interventions had no effect on linear growth (Fig. 2). The high validity of these studies and the consistent effects across three separate sites constitute good quality evidence that these basic WASH interventions, as delivered in these settings, did not reduce stunting. In addition, the novel factorial design of these trials provides good evidence that these WASH interventions in these populations offered no additonal benefit to the evaluated nutrient supplementation intervention as delivered alone. Whilst the effects on linear growth were consistent the underlying reasons for this lack of effect may differ between settings.

The observed effects on diarrhea were mixed, ranging from no effect in Kenya [2] and Zimbabwe [3], to a large relative risk reduction in Bangladesh – albeit against a much lower baseline prevalence [1]. These differences could be the result of interactions between the interventions and features of the study setting and/or population; for example, the local etiologies of diarrheal disease, pre-intervention WASH conditions, the relative importance of different environmental transmission pathways for diarrhea, and the relative importance of zoonotic agents of infection. Recent studies demonstrate the diversity of diarrheal disease etiology across settings and age groups [11, 12], with the transmission of different pathogens more or less likely to be interrupted by basic WASH interventions. For example, Cryptosporidium, a well-established waterborne cause of both endemic [11] and epidemic diarrhea [13], is highly chlorine resistant, thereby likely rendering chlorination, as evaluated in these trials, ineffective [14].

As pointed out by numerous researchers over the decades, different environmental settings require different WASH interventions [15], and the same interventions may even have different effects on health in the same settings at different times [16]. In grossly contaminated environments, where childhood exposure to a variety of enteric pathogens occurs through multiple environmental pathways, partial or even absolute elimination of a single pathway may yield no health benefit. At the same time, under different conditions, small incremental gains may, in some cases, prove catalytic [17, 18]. Alternatively, while some interventions may fail to significantly reduce endemic diarrheal disease, they may still offer protection against epidemic diarrheal disease events [1].

It is well-established that contact with human feces is hazardous to human health: human feces contain various disease-inducing viruses, bacteria, protozoa, and other parasites [19]. Ingestion of these microorganisms in sufficient quantity has been demonstrated to cause disease in decades of challenge studies for a range of pathogens, e.g. Vibrio cholerae [20], Shigella [21], and Campylobacter [22]. Fecal–oral transmission of these pathogens can occur by multiple environmental pathways [23], and all WASH interventions can plausibly prevent some fraction of that transmission. This logic is not challenged by these findings, but the mixed results for diarrhea suggest that these interventions had heterogenous effects on childhood environmental exposure to enteric pathogens [24].

Two of the three trials [1, 2] published ancillary studies to assess the effects of the intervention on environmental contamination [25‐27]. They did this by quantifying fecal indicator bacteria (Escherichia coli) in environmental media corresponding to environmental transmission pathways for diarrheal disease. In Kenya, only the water treatment arm reduced E. coli levels in stored drinking water, and no WASH intervention reduced E. coli levels on children’s hands or on sentinel objects [27]. In Bangladesh, two studies were conducted: the first, four months after the intervention, sampled drinking and ambient water, children’s hands, food given to young children, courtyard soil and flies, in the sanitation only and combined WASH arms [26]; the second, 12 and 24 months after the intervention, sampled drinking water at source and as stored, children’s hands, children’s food and sentinel objects [25]. In the first of these two studies [26], the prevalence of E. coli in stored water was reduced only in the combined WASH arm (prevalence ratio [PR] 0.38; 95% CI:0.32–0.44), with no effect on any other sampled pathway (soil, hygiene, flies or food). In the second study, the prevalence of E. coli in stored drinking water was reduced by the water treatment only intervention (PR 0.62; 95% CI 0.53–0.72) and combined WASH intervention (PR 0.75; 95% CI 0.69–0.81), and the prevalence of E. coli in food was reduced in the single water treatment arm (PR 0.70; 95% CI 0.57–0.86), the single handwashing (PR 0.68; 95% CI 0.56–0.83) and combined WASH interventions (PR 0.89; 95% CI: 0.78–1.01) [25].

Against a low baseline prevalence of diarrhea, and the limited environmental impact of the WASH interventions evaluated, it is notable that a 40% relative reduction in diarrheal disease prevalence was achieved in Bangladesh (an absolute reduction of approximately two percentage points compared to a one-week prevalence of 5.9% in the control arm) [1]. This result was strengthened by a separate comparison of the prevalance of giardiasis across study arms which also showed a marked reduction in infections among all WASH arms except water chlorination [28]. In both the Kenya and Bangladesh trials, which included chlorination only intervention arms, bacterial contamination of stored drinking water was reduced in this arm, but there was no effect on diarrheal disease. As discussed above, this may be due to the resistance of certain diarrhegenic pathogens’ resistance to chlorine, e.g. Cryptosporidium and Giardia [29].

These results suggest that, in settings such as these, more comprehensive or ambitious interventions may be needed to achieve a major impact on child health. In different settings, with more limited WASH conditions – for example, where most people practice open defecation or rely on untreated surface water – these interventions may still yield benefits. Alternatively, in similar settings, more ambitious interventions that address other potentially important exposure sources and/or routes, such as animal waste or foodborne transmission, may be effective in reducing diarrheal disease.

Globally, as countries and regions have transitioned from scenarios in which most of the population have limited or basic WASH services, to one in which most have access to safely managed services, there have been large coincident improvements in public health. Often these improvements have been dramatic with regards to child health and mortality, specifically [30‐32]. They have commonly been associated with major improvements in water and sanitation infrastructure akin to the SDG category of ‘safely managed services’ - that is ensuring a piped supply of safe drinking water directly to the household, or reticulated transportation of human waste to treatment facilities – rather than the more modest changes in service access evaluated in these trials.

Typically, these changes took place over decades. For example, in Victorian Britain, while the great municipal water reforms began in the 1840s, it was not until the 1870s that major investments in sewered household connections began in most cities [32]. Changes in diarrheal disease mortality followed slowly. In London, for example, infantile diarrheal disease mortality was still rising in 1900. In fact, trends in child and infant mortality suggest that benefits accrue from incremental and progressive steps over time, and that major health dividends may come late in that process, as was the case in England and Wales (1900–1920) [32] or in the USA (1920–1930) [31]. Specific innovations are sometimes credited with these health benefits – e.g., in the USA, disinfecting water supplies coincided with major health gains – but crediting these health improvements to water treatment alone ignores, for example, the fact that the infrastructure for the distribution of this treated water was already in place. The lesson perhaps lies in not seeking to attribute benefits to individual WASH factors but in that the public health dividends are paid when comprehensive services are in place, as now envisaged under the new SDG.

These new studies do not challenge the general view that large-scale improvements in water and sanitation infrastructure played an important historical role in improving child health in high-income countries (HIC). The interventions coinciding with dramatic improvements in child health in many of these HICs represented decades of large-scale public investment in piped drinking water and sewered sanitation, as opposed to the provision of basic pit latrines, point-of-use chlorination of water, and handwashing stations as evaluated under these trials.

A Cochrane Review published in 2013, which addressed the effect of WASH interventions on linear growth, identified very few rigorous studies. It concluded there was, “weak evidence of a borderline statistically significant small effect of 0.08 HAZ” [33]. Subsequent intervention studies have produced mixed results, from significant improvements in linear growth [34, 35] to no effect [36‐38]. The new trials considered here hypothesised that WASH interventions might improve linear growth among children by reducing symptomatic and asymptomatic enteric infections, with this effect mediated at least in part by environmental enteric dysfunction (EED) [5, 6], a subclinical condition affecting gut structure and function [39]. The reported effects on symptomatic enteric infections – that is, caregiver-reported diarrheal disease – were mixed, as discussed above, but all three studies found no effect of any WASH intervention arm on linear growth. Furthermore, these factorial studies were specifically designed to assess the combined and independent effects of WASH and nutrition interventions on linear growth, and found no additive benefit of these basic WASH interventions versus nutrition (that is, improved infant and young child feeding, including daily, small-quantity lipid-based nutrient supplements during the period of complementary feeding), alone.

If symptomatic and asymptomatic enteric infection contributes to linear growth faltering, as hypothesized by these studies [5, 6], these results suggest that preventing these infections is unlikely to be achieved with low-cost basic household interventions in highly contaminated settings where young children are exposed to enteric pathogens repeatedly and via multiple routes. To what extent more intensive WASH interventions – such as providing a microbially safe and continuous supply of drinking water piped to the household, or a community-level sewered sanitation system – might impact childhood stunting remains an open question that has so far not been addressed by rigorous trials [33]. The lack of experimental studies of such interventions reflects multiple inherent challenges, including the difficulty of randomly allocating networked infrastructure; the generally high levels of population movement in urban areas; and the long follow-up and large sample sizes required to study linear growth among children. These challenges are not insurmountable, but are certainly formidable.

The results reinforce a well-established view in the nutrition sector that tackling a multifactorial chronic condition such as stunting requires broad and sustained action at multiple levels and across different sectors [40]. The postnatal nutrient supplementation and complementary feeding behaviour change interventions in these same trials, achieved only modest gains in linear growth, despite very high and sustained compliance. These effects are consistent with the wider literature [41, 42], further demonstrating that the underlying causes of stunting remain remarkably poorly understood, and that population-level reductions likely require broad strategies to increase both the availability of nutrients and to reduce nutrient malabsorption. The experience of Brazil, where a dramatic decline in stunting has been achieved, is of three decades of sustained action across multiple sectors, including food, health, social protection as well as water and sanitation [43].

In summary, the basic, household-level interventions evaluated under these trials did not test the ambitious new SDG targets of universal access to safely managed water and sanitation, and therefore do not provide evidence for or against this level of service. The interventions evaluated in these trials effected only modest changes with regard to the ‘WASH ladder’ concept that has been developed for tracking progress against the SDG targets [7]. This ladder represents incremental gains in the quality of WASH services, and there is some evidence to support the assumption that, as the quality of services improve, so too does the magnitude of effect on diarrheal disease. However, the evidence for higher levels of service (e.g., piped drinking water and sewered sanitation connections) is generally of low quality [44].

The ‘sanitation’ interventions included in these trials did not seek to “end open defecation” at a community level, which is a key component of the WASH SDG (SDG 6.2). Instead, the trials focused on improving sanitation at the level of the child’s household, or immediately around the household, as earlier formative work suggested that children’s environmental exposure to enteropathogens occurred within the household or the immediate compound domains [45, 46]. The effect of these interventions on community-level sanitation was, therefore, limited. In fact, at baseline, in both Bangladesh and Kenya, access to household sanitation facilities was relatively high pre-intervention.

The ‘water’ interventions in these trials sought only to improve the microbial quality of drinking water drawn from existing water sources by promoting chlorination in the household. Under the WASH SDG, safely managed drinking water is defined as, “drinking water from an improved water source that is located on premises, available when needed and free from fecal and priority chemical contamination” (SDG 6.1) [7]. In these trials, the distribution of drinking water was not changed to bring water sources closer to the household, to limit service disruptions to ensure water was available when needed, and chemical or microbial contamination of drinking water at source was not addressed. In relation to how water influences disease transmission, an old distinction can be drawn between ‘waterborne’ transmission, that is where transmission occurs via ingestion of water containing pathogens, and ‘water-washed’ transmission, wherein person-to-person transmission results from insufficient water to practice adequate personal and domestic hygiene [47, 48]. It has been observed that when the household water supply is on-plot or piped directly into the household, the amount of water consumed increases dramatically [49], and to a level wherein adequate water is available to meet hygiene needs and reduce health risks [50, 51]. Critically, then, neither the distance to water source nor the volume of water consumed was changed by these interventions.

These trials did not evaluate the effect of safely managed water and community-level safely managed sanitation services, as called for under the new WASH SDG, on child stunting or diarrhea. The basic interventions evaluated in these trials were seemingly insufficient to comprehensively reduce enteric pathogen exposure, and had mixed effects on diarrhea. This may, in part, explain the lack of effect on stunting, if diarrhea mediates at least some part of this relationship.